Detection and volume estimation of bubbles in blood circuit of hemodialysis by morphological image processing

Air bubbles in extracorporeal blood circuit of hemodialysis are monitored and evaluated in case of flowing into the patient. The bubbles in the dialysis device are detected by the optical or ultrasonic devices, and micro bubbles can pass the air sensor without causing system alarm. This paper proposes a morphological method to identify the number and the cumulative volume of the bubbles accurately. The gray morphology is used for image preprocessing. After preprocessing, the image is filtered by binary morphological method. Then methods for estimation of bubble volume are applied with the mural bubble and air embolus being equivalent to cylinder and ellipsoid respectively. The results of the detection and the calculation are relatively accurate, the absolute errors of cumulative volumes are within the acceptable level.